#!/bin/python
import boto3
import sys
import os
import argparse
import logging
from multiprocessing import Process


def parse_arguments():
    parser = argparse.ArgumentParser()
    parser.add_argument('--profile', help='AWS profile', default='gdb-delphi-prod')
    parser.add_argument('--concurrency', type=int, help='Size of a process pool', default=os.cpu_count())
    requiredNamed = parser.add_argument_group('required named arguments')
    requiredNamed.add_argument('--bucket', help='s3 bucket name', required=True)
    requiredNamed.add_argument('--canonicalId', help='Canonical ID of bucket owner user', required=True)
    requiredNamed.add_argument('--keyslistfile', help='file with list of keys', required=True)
    requiredNamed.add_argument('--outfile', help='file with list of wrong objects', required=True)
    return parser.parse_args()

def get_object_if_wrong_owner(client,bucket,key,canonicalId):
    acl = client.get_object_acl(Bucket=bucket,Key=key)    
    if (acl['Owner']['ID'] != canonicalId):
        return f's3://{bucket}/{key}\n'
    else:
        return ''
        
        

def process_object_list(profile,bucket,keylist,canonicalId,filename):
    session = boto3.Session(profile_name=profile)
    client = session.client('s3')
    object_list = ''
    for key in keylist:
        object_list += get_object_if_wrong_owner(client,bucket,key,canonicalId)
    with open(filename, 'a') as fd:
        fd.write(object_list)

def partitioned_keys_list(keys_list, partitions_num):
    chunk_size = int(len(keys_list)/partitions_num)
    for i in range(0, len(keys_list), chunk_size):
        if ((i + chunk_size) < len(keys_list)):
            yield keys_list[i:(i + chunk_size)]
        else:
            yield keys_list[i:]

if __name__ == "__main__":
    logger = logging.getLogger('')
    logger.setLevel(logging.DEBUG)

    args = parse_arguments()

    # I used aws cli for creating keys list
    # aws s3api list-objects --profile debug_profile --bucket bucket_name --query "(Contents)[].[Key]" --output text
    keys_list = [line.rstrip('\n') for line in open(args.keyslistfile, 'r')]
    #print(keys_list)
    #process_object_list(s3_client,args.bucket,keys_list,args.canonicalId,args.outfile)    
    for partitioned_list in partitioned_keys_list(keys_list, args.concurrency):
        print(len(partitioned_list))
        p = Process(target=process_object_list, args=(args.profile,args.bucket,partitioned_list,args.canonicalId,args.outfile))
        p.start()